From 61f2e8d45b4e43f7aa898a1ba0f872beccb9775c Mon Sep 17 00:00:00 2001
From: Feotov Daniil <hairyhum@gmail.com>
Date: Mon, 7 Apr 2014 17:38:11 +0400
Subject: [PATCH 1/4] Fixed error in ucs_to_utf

---
 c_src/jsonx_str.h  | 2 +-
 test/str_tests.erl | 2 ++
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/c_src/jsonx_str.h b/c_src/jsonx_str.h
index 606eb8d..50dde5d 100644
--- a/c_src/jsonx_str.h
+++ b/c_src/jsonx_str.h
@@ -103,7 +103,7 @@ ucs_to_utf8(unsigned char* ptr, unsigned ucs){
     *(ptr++) = (unsigned char) 0xC0 + (ucs >> 6);
     *(ptr++) = (unsigned char) 0x80 + (ucs & 0x3F);
     return ptr;
-  }else if(ucs < 0x1000) {
+  }else if(ucs < 0x10000) {
     // 1110xxxx 10xyyyyy 10yyyyyy
     if(ucs < 0xD800 || (ucs > 0xDFFF && ucs < 0xFFFE)) {
       *(ptr++) = (unsigned char) 0xE0 + (ucs >> 12);
diff --git a/test/str_tests.erl b/test/str_tests.erl
index ba8eac1..36aa7df 100644
--- a/test/str_tests.erl
+++ b/test/str_tests.erl
@@ -33,6 +33,8 @@ encutf4_test() ->
 encutf5_test() ->
      {no_match,<<248,128,128,128,128>>} = jsonx:encode(<<248, 128, 128, 128, 128>>).
 
+decutf_test() ->
+    <<237,129,172,235,166,176,236,138,164,237,139,177>> = jsonx:decode(<<"\"\\ud06c\\ub9b0\\uc2a4\\ud2f1\"">>).
 
 %% Test decode atoms
 dectrue_test() ->

From d5b22478871d5e8e5afeba90ac425f70d8b41f81 Mon Sep 17 00:00:00 2001
From: Feotov Daniil <hairyhum@gmail.com>
Date: Thu, 10 Apr 2014 11:52:10 +0400
Subject: [PATCH 2/4] Parse utf16 surrogate pairs

---
 c_src/jsonx_str.h | 37 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 36 insertions(+), 1 deletion(-)

diff --git a/c_src/jsonx_str.h b/c_src/jsonx_str.h
index 50dde5d..15a5d31 100644
--- a/c_src/jsonx_str.h
+++ b/c_src/jsonx_str.h
@@ -124,6 +124,35 @@ ucs_to_utf8(unsigned char* ptr, unsigned ucs){
   return NULL;
 }
 
+static inline int
+is_surrogate_start(int val){
+  return ((0xD800 <= val) && (val <= 0xDBFF));
+}
+
+static inline int
+next_surrogate_end(unsigned char *str){
+  unsigned char *src = str;
+  src+=4;
+  if(*src == '\\'){
+    unsigned hval;
+    src++;
+    if(*src == 'u'){
+      src++;
+      if(ucs_from_4hex(src, &hval)) {
+        if((0xDC00 <= hval) && (hval <= 0xDFFF)){
+          return hval;
+        }
+      }
+    }
+  }
+  return 0;
+}
+
+static inline int
+parse_surrogate(int start, int end){
+  return (((start - 0xD800) << 10) + (end - 0xDC00) + 0x0010000);
+}
+
 static inline  int
 check_with_unescape_jstr(unsigned char *str, unsigned char **endstr, unsigned char **endptr){
   unsigned char c, k;
@@ -174,8 +203,14 @@ check_with_unescape_jstr(unsigned char *str, unsigned char **endstr, unsigned ch
       case '\\': {src++; *dst++ = 92U; continue;}
       case 'u': {
 	unsigned hval;
+  src++;
+  if(!ucs_from_4hex(src, &hval)) {goto error;}
+  unsigned sur_end;
+  if(is_surrogate_start(hval) && (sur_end = next_surrogate_end(src))){
+    hval = parse_surrogate(hval, sur_end);
+    src += 6;
+  } else {goto error;}
 	src++;
-	if(!ucs_from_4hex(src, &hval)) {goto error;}
 	if(!(dst = ucs_to_utf8(dst, hval))) {goto error;}
 	src += 4;
 	continue;

From 7f8138b1484326f85da2c06978560eb32770298f Mon Sep 17 00:00:00 2001
From: Feotov Daniil <hairyhum@gmail.com>
Date: Thu, 10 Apr 2014 12:35:27 +0400
Subject: [PATCH 3/4] Fix erorr in parsing surrogate pairs

---
 c_src/jsonx_str.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/c_src/jsonx_str.h b/c_src/jsonx_str.h
index 15a5d31..88fc5cb 100644
--- a/c_src/jsonx_str.h
+++ b/c_src/jsonx_str.h
@@ -210,7 +210,6 @@ check_with_unescape_jstr(unsigned char *str, unsigned char **endstr, unsigned ch
     hval = parse_surrogate(hval, sur_end);
     src += 6;
   } else {goto error;}
-	src++;
 	if(!(dst = ucs_to_utf8(dst, hval))) {goto error;}
 	src += 4;
 	continue;

From 7044db6bebcae2bd4691f925f80b7687555c926f Mon Sep 17 00:00:00 2001
From: Feotov Daniil <hairyhum@gmail.com>
Date: Thu, 10 Apr 2014 14:16:55 +0400
Subject: [PATCH 4/4] Do not return error if not surrogate

---
 c_src/jsonx_str.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/c_src/jsonx_str.h b/c_src/jsonx_str.h
index 88fc5cb..4a7fd33 100644
--- a/c_src/jsonx_str.h
+++ b/c_src/jsonx_str.h
@@ -209,7 +209,7 @@ check_with_unescape_jstr(unsigned char *str, unsigned char **endstr, unsigned ch
   if(is_surrogate_start(hval) && (sur_end = next_surrogate_end(src))){
     hval = parse_surrogate(hval, sur_end);
     src += 6;
-  } else {goto error;}
+  }
 	if(!(dst = ucs_to_utf8(dst, hval))) {goto error;}
 	src += 4;
 	continue;